PRE-SEASON RUN SIZE FORECASTS FOR FRASER RIVER Canadian Science Advisory Secretariat

advertisement
Pacific Region
Canadian Science Advisory Secretariat
Science Advisory Report 2009/022
PRE-SEASON RUN SIZE FORECASTS FOR FRASER RIVER
SOCKEYE AND PINK SALMON IN 2009
Figure 1. Sockeye adult spawning phase (source: Figure 2. Pink adult spawning phase (source: DFO
DFO website) and Fraser watershed distribution
website) and Fraser watershed distribution (DFO
(DFO GIS Division).
GIS Division).
Context
Pre-season abundance forecasts of returning Fraser River adult sockeye and pink salmon in 2009 were
requested by Fisheries and Oceans Canada (DFO) Fisheries Management. Forecasts are used for preseason planning purposes and for in-season management. They are most useful early in the summer
fishing season before reliance on in-season run size estimates. Forecasts are produced by DFO as
agreed under the United States (U.S.)-Canada Pacific Salmon Treaty. Detailed methods are
documented in Cass et al. (2006) and DFO (2007). Forecasts have been reviewed annually and are
publicly available: http://www.meds-sdmm.dfompo.gc.ca/csas/applications/Publications/publicationIndex_e.asp
SUMMARY
!
The 2009 forecast of sockeye returns at the 50% probability level for all 19 stocks plus
miscellaneous stocks is 10.6 million fish (255,000 Early Stuart, 739,000 Early Summer,
and 8.7 million Summer and 907,000 Late).
!
This 50% probability level return forecast (10.2 million excluding miscellaneous stocks)
is below the long-term average for this cycle of 13 million fish (1980-2005). The
Summer Run return forecast (48% Chilko & 41% Quesnel) accounts for 82% of the total
forecast. Of the total forecast, Late Run stocks comprise 9%, Early Summer Run stocks
7% and the Early Stuart Run stock comprise 2%.
May 2009
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
!
The 2009 forecast of pink salmon returns is 17.5 million (50% probability level).
!
Forecasts are associated with relatively high uncertainty, consistent with previous Fraser
sockeye forecast PSARC reviews (Cass et al. 2006) and recent research on coast-wide
salmon stocks (Haeseker et al. 2007 & 2008).
!
For adult sockeye returning in 2009, ocean conditions in their juvenile outmigration
period (spring/summer 2007) were good for most physical (sea surface temperatures,
upwelling, spring transition) and biological (zooplankton species composition &
abundance) indicators. This was an improvement over the transitional year in 2006 and
the poor year in 2005 associated with reduced ocean survival and relatively poor returns
in 2007 and 2008 for BC salmon stocks. Given the positive ocean productivity signals in
2007 relative to 2006 (transitional between poor and good) & 2005 (poor), emphasis on
median probability levels (50%) is recommended for 2009 sockeye return forecasts with
the exception of Early Stuart.
!
The forecasts for the Early Stuart power model at the 75% probability level for the past
five years were closer to true returns compared to forecasts at the 50% probability level.
INTRODUCTION
The Summer Run timing group that includes Quesnel, Late Stuart, Chilko, and Stellako are
drivers of the 2009 cycle return abundances accounting for respectively, 54%, 17%, 11%, and
3% of the total sockeye returns on the cycle since 1980. The 2009 cycle is a dominant cycle
year for Quesnel and Late Stuart stocks. On the 2009 cycle, average annual returns for all 19
forecasted stocks combined were 13 million (Table 1).
Most Fraser sockeye are comprised of predominantly age-4 fish (Gilbert-Rich aging convention:
42), spending their first two winters in freshwater and last two winters in the marine environment
prior to returning to the freshwater to spawn. Therefore, most sockeye that return in 2009 are
recruited from eggs spawned by adults in 2005 (brood year). In the 2005 brood year, the
number of effective female spawners (product of the number of female spawners and the
proportion that successfully spawned) for most of the 19 forecasted Fraser sockeye stocks was
close to average or above average (time series: 1980-2005) with the exception of Early Stuart,
Bowron, Seymour, Late Stuart, Quesnel, & Birkenhead (Table 3). The greatest contributors to
the 2005 brood year escapements (82% of the total) were Quesnel (45%), Chilko (16%),
Harrison (12%), and Late Stuart (9%). Each of the other 15 forecasted stocks contributed
approximately 5% or less to the total. Most sockeye stocks have an age-5 (52) component and
for most of these stocks the numbers of effective female spawners contributing to age-5 returns
in 2009 (2004 brood year) were below average (Table 3).
From the 2005 brood year spawners, juvenile production relative to the 1980-2005 time series
varied between stocks. Quesnel fry abundance assessed in 2006 (52 million fish) was below
average (58 million), consistent with the lower than average number of effective female
spawners in the 2005 brood year. For Cultus sockeye, although the number of effective female
spawners was particularly low in the 2005 brood year, hatchery enhancement activities for this
stock produced 100,000 smolts in 2007 which is double the time series average of 50,000
(Table 3). Chilko sockeye smolt numbers that outmigrated in 2007 (77 million), produced from
2
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
2005 spawners, were the highest number of smolts recorded for this stock (1980-2005 average:
23 million) (Table 3). Chilko smolt body sizes were also relatively large in 2007 (88 mm versus
82 mm average from 1980-2003).
Return forecasts are made for each of 19 individual sockeye stocks that have historical
escapement and return data (Table 1). Together the 19 sockeye stocks accounted for 97% of
the 2005 brood year escapement to the Fraser River.
Forecasts for the remaining
miscellaneous stocks (escapement data only available) accounted for 3% of the 2005 brood
year escapement.
Forecasts of sockeye returns for the 19 stocks are typically made using a variety of methods
that include naïve and biological models. Model selection for each stock depends on data
availability and model performance using retrospective analysis (Cass et al. 2006). Uncertainty
in sockeye forecasts for 2009 is captured using Bayesian statistical inference. Miscellaneous
stocks have only escapement data (return data are not available) and these are forecast as a
product of their brood year effective female spawners and average recruits per spawners for
their respective run timing groups. Sockeye forecasts presented here are based on the same
methods and data streams reported in Cass et al. (2006) and DFO (2007) except for the
addition of recent years data required to generate 2009 forecasts.
Fraser Sockeye Indicator Stock: Chilko
Chilko River sockeye is the only stock for Fraser sockeye with a complete adult and juvenile
time series (1949-present) obtained from relatively high accuracy & precision assessments
(mark recapture analyses of adults and estimates of outmigrating juveniles at a weir on Chilko
River). Chilko is unlikely to provide an indication of freshwater survival for all Fraser sockeye
stocks given the broad differences in lake characteristics (e.g. hydrology, limnology, food web)
that occur between systems. Chilko, however, may be an indicator of marine survival for Fraser
sockeye stocks depending on their similarities in migration timing and ocean distributions.
For Chilko, the highest freshwater survival observed for this stock occurred in the past two
brood years (2004 and 2005: 6%) compared to the long term average (brood years 1949-2003:
3%) (Figure 3). Smolt body sizes have also been large in the past two brood years (2004: 100
mm & 2005: 88 mm) relative to the long term average (82 cm). No limnological or food web
dynamic surveys have occurred in recent years in Chilko Lake but it is appears there has been
a shift in productivity specific to this system (J. Hume, pers. comm.). In contrast, average
marine survival has been relatively low in recent years (brood years 2000-2003: ~3%)
compared to the long term average (brood years 1949-2003: ~9%) (Figure 3). Ocean
conditions were particularly poor in 2005 resulting in low sockeye returns in 2007.
3
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Figure 3. Chilko sockeye freshwater and marine survival.
Methods
Data Sources and Methods:
Data sources and methods have been extensively reviewed by Pacific Scientific Advice
Review Committee (PSARC) and are presented in Cass et al. (2006) and DFO (2007)
available on the Canadian Science Advisory Secretariat (CSAS) website:
http://www.meds-sdmm.dfo-mpo.gc.ca/csas/applications/Publications/publicationIndex_e.asp
Forecast Models
Forecast model descriptions are presented in detail by Cass et al. (2006):
http://www.dfo-mpo.gc.ca/csas/Csas/Publications/ResDocs-DocRech/2006/2006_060_e.htm
& DFO (2007): http://www.pac.dfo-mpo.gc.ca/sci/psarc/SSRs/Salmon/sar2007-049_e.pdf.
Bayesian prior probability distributions for the biological model parameters for forecast model
are presented Cass et al. 2006 (Appendix 3) and DFO (2007).
Retrospective Analysis
Retrospective analyses were conducted based on methodology described in Cass et al. (2006).
Two performance measures were used to rank each model’s performance: mean absolute error
(MAE) and root mean square error (RMSE).
Smaller differences or errors between the
forecasted number of returning salmon and the true return number indicate better model
performance. Each model was ranked based on how it performed for each of these two
performance measures and the two ranks were then averaged together to produce a final
ranking (Table 4).
Forecasts for 2009 were generated for the top three ranking models. If a stock also had
juvenile data available, forecasts were generated for top ranking effective female spawner and
4
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
juvenile biological models. Results for the top ranking models are considered similar unless,
the median forecast of one model is more extreme than the 25% or 75% probability level
forecast of another model. In cases where results for models were similar, the top ranked
model was used to provide the forecast estimates. In cases where results differ between naïve
and biological models, biological models get preference. In cases where no additional
information exists to evaluate the validity of the different models, divergent estimates were
averaged (weighted based on retrospective analysis performance) (Fried & Yuen 1987).
ASSESSMENT
Forecasts based on the best candidate model are provided at various probability levels of
achieving specified run sizes by stock and run-timing group (Table 1). The 2009 forecast of
sockeye returns at the 50% probability level for all 19 stocks plus miscellaneous stocks is 10.6
million fish (255,000 Early Stuart, 739,000 Early Summer, 8.7 million Summer and 907,000
Late). This 50% probability level return forecast (10.2 million excluding miscellaneous stocks) is
below the long-term average for this cycle of 13 million fish (1980-2005). The Summer Run
return forecast (48% Chilko & 41% Quesnel) accounts for 82% of the total forecast. Of the total
forecast, Late Run stocks comprise 9%, Early Summer Run stocks comprise 7% and the Early
Stuart Run stock comprises 2%. The 2009 forecast of pink salmon returns at the 50%
probability level is 17.5 million fish.
5
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Table 1. Pre-season sockeye and pink forecasts for 2009 by stock/timing group and probability.
Biological model predictor variable indicated (i.e. fry, smolt, or effective female spawners: eff). The Wild
Salmon Policy (WSP) conservation units (CU’s) that each forecasted stock comprises is numerically
referenced below (1 to 32), with corresponding CU name and index listed in Table 2.
Probability of Achieving Specified Run Sizes
Sockeye stock/timing
CU's
b
group
(Table 2) Forecast Model
d
Early Stuart
1,2
Pooled
Early Summer
(total exlcuding miscellaneous)
Bowron
Fennell
Gates
Nadina
Pitt
Raft
Scotch
Seymour
e
Misc
f
Misc
g
Misc
h
Misc
i
Misc
Summer
Chilko
Late Stuart
j
Quesnel
Stellako
3
4
5
6,7
8
4
9
9
9
4,10
11
12
4
Ricker (eff)-pi
Ricker (eff)
Ricker (eff)-cyc
Ricker (eff)-peak
Power (eff)
Power (eff)
RS1 (naïve)
Ricker (eff)-cyc
R/S
R/S
R/S
R/S
R/S
13,14
15,16
17,18,19
19,20,21
Power (smolt)
R1C (naïve)
Pooled
Larkin (eff)
Late
(total exlcuding miscellaneous)
Cultus
Harrison
Late Shuswap
Portage
Weaver
Birkenhead
k
Misc. Shuswap
k
Misc. non-Shuswap
22
23
24
25
26
27
24,28,29
30,31
TOTAL
(TOTAL excluding miscellaneous)
Pink Salmon
32
Power (Smolt)-Jack
Ricker (eff)-PDO
Ricker (eff)-cyc
Ricker (eff)
Larkin (eff)
Power (eff)
R/S
R/S
Mean Run Size
a
c
all cycles
2009 cycle
335,000
797,000
0.1
645,000
0.25
426,000
0.5
255,000
0.75
165,000
0.9
107,000
-
-
2,284,000
1,338,000
739,000
443,000
264,000
(501,000)
(316,000)
(1,234,000)
(749,000)
(443,000)
(272,000)
(177,000)
23,000
28,000
65,000
81,000
61,000
32,000
64,000
147,000
-
13,000
17,000
52,000
78,000
79,000
31,000
20,000
26,000
-
-
-
25,000
101,000
224,000
181,000
270,000
209,000
170,000
54,000
22,000
47,000
25,000
156,000
800,000
16,000
60,000
127,000
118,000
189,000
131,000
77,000
31,000
12,000
27,000
14,000
88,000
448,000
10,000
34,000
74,000
73,000
124,000
78,000
32,000
18,000
6,000
13,000
7,000
44,000
226,000
6,000
21,000
44,000
43,000
86,000
49,000
13,000
10,000
4,000
8,000
4,000
25,000
130,000
4,000
12,000
30,000
28,000
58,000
32,000
6,000
7,000
2,000
4,000
2,000
13,000
66,000
11,111,000 31,813,000 16,071,000
1,396,000 9,466,000 6,136,000
2,300,000 3,538,000 1,469,000
7,082,000 18,037,000 7,936,000
333,000
772,000
530,000
8,677,000
4,175,000
553,000
3,575,000
374,000
4,914,000
2,870,000
208,000
1,575,000
261,000
2,858,000
1,857,000
86,000
724,000
191,000
5,677,000
1,760,000
834,000
2,556,000
527,000
-
-
2,875,000
1,616,000
907,000
517,000
327,000
(3,242,000)
(946,000)
(2,665,000)
(1,482,000)
(843,000)
(485,000)
(310,000)
19,000
47,000
2,204,000
58,000
432,000
482,000
-
3,000
NA
78,000
74,000
332,000
459,000
-
16,000
373,000
407,000
259,000
906,000
704,000
91,000
119,000
10,000
160,000
171,000
140,000
546,000
455,000
56,000
78,000
5,000
69,000
70,000
66,000
336,000
297,000
27,000
37,000
3,000
46,000
26,000
31,000
200,000
179,000
17,000
15,000
1,000
33,000
10,000
16,000
126,000
124,000
11,000
6,000
-
- 37,617,000 19,451,000 10,578,000
6,039,000
3,556,000
(9,755,000)
(13,170,000) (36,357,000) (18,728,000) (10,218,000)
(5,836,000)
(3,452,000)
- 32,939,000 24,858,000 17,535,000 12,490,000
9,343,000
12,067,000
a. probability that the actual run size will exceed the specified projection.
b. see Cass et al. (2006) and DFO (2007) for model descriptions.
c. sockeye: 1980-2005 (excluding miscellaneous stocks); pink: 1961-2005.
d. Early Stuart is pooled power and RS2 model (average weighted from retro analysis)
e. unforecasted misc. Early Summer stocks (Early Shuswap stocks: S.Thompson); return timing most similar to Scotch/Seymour).
f. unforecasted misc. Early Summer stocks (N. Thomson tributaries; return timing most similar to Fennell/Bowron/Nadina).
g. Nahatlach River & Lake
h. Chilliwack Lake and Dolly Varden Creek; return timing most similar to Early Stuart.
i. North Thompson River.
j. Quesnel is a pooled Larkin and Power model (average weighted from retro analysis performance during dominant yr)
k. unforecasted miscellaneous Late Run stocks; true lates made up a very small component (~800 at 50% prob. level)
Model definitions: pi (Pine Island SST covariate); cyc (cycle line data only); peak (Fraser R. peak discharge covariate); PDO
(Pacific Decadal Oscillation (PDO) covariate); RS1 (product of R/S from last generation & eff fem spawners in brood year);
RS2 (product of R/S from last 2 generations and eff fem spawners in brood year); R1C (rec from last generation); R/S (used
for stocks with no recruit data: product of R/S for run timing group and eff fem spawners).
6
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Table 2. Conservation units’ (CU’s) names and index’s referenced in Table 1 for each forecasted stock
group, including miscellaneous stocks, are listed below (Holtby & Ciruna, 2007).
CU Number
(Table 1)
CU Name
CU Index
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Stuart-EStu
Takla/Trembleur-Estu
Bowron-ES
Kamloops-ES
Anderson-ES
Francois-ES
Nadina-ES
Pitt-ES
Shuswap Complex-ES
Taseko-ES
Nahatlach-ES
Chilliwack-ES
Chilko-ES
Chilko-S
Stuart-S
Takla/Trembleur-S
McKinley-S
Quesnel-S
MFR
Fraser-S
Francois-S
Cultus-L
LFR
Shuswap Complex-L
Seton-L
Harrison (U/S)-L
Lillooet-L
Carpenter-L
Kamloops-L
Harrison (D/S)-L
Widgeon
L-06-12
L06-14
L-07-01
L-10-01
L-06-01
L-06-04
L-06-09
L-03-05
L-09-02
L-06-16
L-05-02
L-03-01
L-06-02
L-06-03
L-06-13
L-06-15
L-06-08
L-06-10
R05
L-06-07
L-06-05
L-03-02
R03
L-09-03
L-06-11
L-03-04
L-04-01
L-06-17
L-09-01
L-03-03
R02
32
Fraser River
FR
Sockeye
Pink
7
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Table 3. Brood year effective female spawners and/or juvenile outmigration (Early Stuart, Chilko,
Quesnel, & Cultus) for Fraser sockeye returning as age-4’s (2005 brood year) and age-5’s (2004 brood
year) in 2009 and the 2009 50% probability forecast (Table 1). Colors indicate escapements and
forecasted returns relative to cycle averages: red (R) indicates below average, yellow (Y) indicates
average, and green (G) indicates above cycle average. BY: brood year.
Stock Name
Early Stuart
Early Sum mer
Bowron
Fennel
Gates
Nadina
Pitt
Raft
Scotch
Seym our
Summer
a
Chilko
Late Stuart
Quesnel
Stellako
Late
Cultus
a
Harrison
b
Late Shuswap
Portage
W eaver
Birkenhead
2005 BY
(age-4)
2004 BY
(age-5)
2009
Forecast
R
R
R
R
R
R
Y
R
G
G
R
G
G
Y
Y
G
G
G
G
R
G
Y
R
G
R
R
R
G
R
G
R
Y
R
R
Y
R
G
R
G
G
Y
G
G
NA
G
G
Y
Y
G
Y
Y
G
R
G
R
R
R
a. Juvenile data for forecasts with juvenile m odels (Early Stuart, Chilko, Quesnel, and Cultus).
b. Harrison sockeye are 3 & 4 year old sockeye (negligible age-5 com ponent).
8
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Table 4. Top ranked models and associated performance measures (MAE & RMSE) for each of the 19
forecasted stocks; MAE: mean absolute error; RMSE: root mean square error. Highlighted (yellow)
models were ruled out (see text under stock’s associated run timing group for explanation). Biological
model predictor variable indicated (i.e. fry, smolt, or effective female spawners: eff).
RUN TIMING: EARLY STUART
EARLY STUART MAE
rank
Power (eff)
0.144 1
RS2 (eff)
0.194 3
Ricker (eff)-ei
0.207 5
RMSE
0.189
0.308
0.328
rank
1
2
4
MEAN
1.0
3.0
4.0
RANK
1
3
4
RUN TIMING: SUMMER
CHILKO
MAE
Power (smolt)
0.781
Power (smolt)-ei
0.826
Larkin (eff)
0.883
rank
1
2
6
RMSE
1.181
1.228
1.101
rank
2
3
1
MEAN
1.5
2.5
3.5
RANK
1
2
3
RUN TIMING: EARLY SUMMER
BOWRON
MAE
rank
Ricker (eff)-pi
0.018 1
Power (eff)
0.02
4
Larkin (eff)
0.021 5
RMSE
0.026
0.028
0.027
rank
1
3
2
MEAN
1.0
3.5
3.5
RANK
1
2
2
LATE STUART
R1C (naïve)
Power (eff)
RAC (naïve)
R2C (naïve)
MAE
0.504
0.565
0.579
0.577
rank
1
2
4
3
RMSE
0.901
1.032
1.057
1.1
rank
1
2
3
4
MEAN
1.0
2.0
3.5
3.5
RANK
1
2
3
3
FENNEL
Ricker (eff)
TSA (naïve)
Ricker (eff)-pi
MAE
0.02
0.02
0.02
rank
1
1
1
RMSE
0.023
0.024
0.026
rank
1
2
3
MEAN
1.0
1.5
2.0
RANK
1
2
3
GATES
Ricker (eff)-cyc
Larkin (eff)
Power (eff)
MAE
0.003
0.004
0.043
rank
1
2
5
RMSE
0.003
0.004
0.058
rank
1
2
3
MEAN
1.0
2.0
4.0
RANK
1
2
3
QUESNEL
Power (fry)-pi
Power (fry)-ei
RAC (naïve)
Larkin (eff)
Power (fry)
MAE
1.24
1.786
1.864
1.894
1.98
rank
1
2
3
4
6
RMSE
1.474
2.12
2.464
3.444
2.603
rank
1
2
3
8
6
MEAN
1.0
2.0
3.0
6.0
6.0
RANK
1
2
3
6
6
NADINA
Ricker (eff)-peak
Ricker (eff)-pi
Power (eff)
Power (fry)-pi
MAE
0.06
0.06
0.06
0.06
rank
1
1
1
1
RMSE
0.115
0.118
0.118
0.125
rank
1
4
4
6
MEAN
1.0
2.5
2.5
3.5
RANK
1
2
2
4
STELLAKO
Ricker (eff)-cyc
Larkin (eff)
R2C (naïve)
MAE
0.048
0.23
0.242
rank
1
2
3
RMSE
0.074
0.282
0.344
rank
1
2
3
MEAN
1.0
2.0
3.0
RANK
1
2
3
PITT
Power (eff)
MRS (naïve)
R2C (naïve)
Ricker (eff)-disch
MAE
0.044
0.048
0.051
0.05
rank
1
2
5
3
RMSE
0.073
0.068
0.088
0.091
rank
2
1
4
6
MEAN
1.5
1.5
4.5
4.5
RANK
1
1
3
3
RUN TIMING: LATE
CULTUS
MAE
Smolt-jack
0.01
RSC (naïve)
0.011
Power (smolt)-peak 0.011
Power (smolt)-ei
0.011
Power (smolt)-pi
0.011
Power (smolt)-PDO 0.011
Power (smolt)
0.011
rank
2
2
2
2
2
2
2
RMSE
0.014
0.018
0.02
0.02
0.02
0.02
0.02
rank
1
2
3
3
3
3
3
MEAN
1.5
2.0
2.5
2.5
2.5
2.5
2.5
RANK
1
2
3
3
3
3
3
RAFT
Power (eff)
Ricker (eff)
R1C (naïve)
MAE
0.013
0.015
0.015
rank
1
2
2
RMSE
0.016
0.02
0.021
rank
1
2
3
MEAN
1.0
2.0
2.5
RANK
1
2
3
SCOTCH
RS1 (naïve)
Ricker (eff)-pi
Ricker (eff)-disch
MAE
0.039
0.04
0.044
rank
1
2
3
RMSE
0.054
0.058
0.082
rank
1
2
3
MEAN
1.0
2.0
3.0
RANK
1
2
3
HARRISON
MRS (naïve)
R1C (naïve)
R2C (naïve)
Ricker (eff)-PDO
Ricker (eff)-disch
Power (eff)
MAE
0.039
0.04
0.04
0.037
0.04
0.042
rank
2
3
3
1
3
7
RMSE
0.077
0.079
0.084
0.086
0.086
0.086
rank
1
2
3
6
6
6
MEAN
1.5
2.5
3.0
3.5
4.5
6.5
RANK
1
2
3
4
5
6
SEYMOUR
Ricker (eff)-cyc
Larkin (eff)
RAC
MAE
0.048
0.067
0.083
rank
1
2
3
RMSE
0.074
0.087
0.144
rank
1
2
3
MEAN
1.0
2.0
3.0
RANK
1
2
3
LATE SHUSWAP
Ricker (eff)-cyc
RAC (naïve)
Ricker (eff)-peak
Ricker (eff)
MAE
0.817
0.736
0.845
0.891
rank
2
1
3
5
RMSE
1.351
1.472
1.469
1.543
rank
1
3
2
4
MEAN
1.5
2.0
2.5
4.5
RANK
1
2
3
4
PORTAGE
Ricker (eff)-cyc
Ricker (eff)
Power (eff)
MAE
0.034
0.038
0.038
rank
1
2
2
RMSE
0.038
0.053
0.054
rank
1
2
3
MEAN
1.0
2.0
2.5
RANK
1
2
3
WEAVER
Larkin (eff)
Ricker (eff)-cyc
Power (fry)-PDO
MAE
0.083
0.127
0.18
rank
1
2
3
RMSE
0.086
0.178
0.239
rank
1
2
3
MEAN
1.0
3.0
4.0
RANK
1
2
3
BIRKENHEAD
Power (eff)
Ricker (eff)-cyc
Larkin (eff)
MAE
0.278
0.314
0.303
rank
1
3
2
RMSE
0.4
0.4
0.414
rank
2
1
3
MEAN
1.5
2.0
2.5
RANK
1
2
3
Early Stuart Sockeye
The 2009 cycle line is the dominant cycle for the Early Stuart Run. Escapement in the 2005
brood year was 51,000 effective females, which is 36% of the cycle average (140,000 from
1980-2005) and the lowest Early Stuart Run escapement on this cycle year in 4 decades (Table
9
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
3). Spawning success in the brood year was high; physical conditions (water levels and
temperature) on the spawning grounds were conducive to successful spawning.
Fry abundance is estimated in the Early Stuart system, however, due to inconsistencies in data
collection further evaluation of this data is required for purposes other than providing an index of
fry abundance. Therefore, fry data was not used to generate forecasts for 2009.
Using effective female spawner data, forecasts generated for the top two ranked models
diverged (power and RS2: average recruits per spawner for the last two generations on the
cycle); the median forecast of the top ranked model was more extreme than the 25% or 75%
probability level forecast of the second ranked model. Given forecasts for these two models
diverged, they were averaged (weighted by retrospective performance). Note: the third ranked
model (Ricker with the Pacific Decadal Oscillation as a covariate) produced a forecast similar to
the first ranked model.
The 2009 median (50% probability level) return forecast for Early Stuart Run sockeye is
255,000. Based on the 2009 forecast distribution there is a 25% probability the number of
returning sockeye will exceed 426,000 sockeye and a 75% probability that the return will exceed
165,000 sockeye (Table 1). The 2009 forecast (50% probability forecast: 255,000) is below the
long term average return for this cycle (1980-2005: 797,000). When comparing the past five
year’s forecasted returns with actual returns, the power model consistently overestimated
returns at the 50% probability level. Forecasted returns for the power model at the 75%
probability level were closer to actual returns for Early Stuart.
Early Summer Run Sockeye
The Early Summer Run consists of several small stocks. Eight stocks in this timing group have
individual forecasts: Bowron, Fennell, Gates, Nadina, Pitt, Raft, Scotch, and Seymour (Table 1).
Escapement in the 2005 brood year was 80,000 effective female spawners for these eight
stocks which is above the cycle average of 52,000 (1980-2005). The total effective female
spawners for the Early Summer Run including the miscellaneous stocks are 130,000. The
North Thompson miscellaneous stock comprises the greatest percentage of this total
escapement at 36%, followed by Pitt (25%), Raft (13%), and Nadina (10%).
For all eight stocks except Bowron and Seymour (both ~60% of the brood year effective female
spawners), the effective number of female spawners in the brood year is similar (Fennell and
Scotch) or greater (Gates, Nadina, Pitt, Raft) than the cycle average (1980-2005) (Table 3). For
Nadina, the only Early Summer stock for which consistent juvenile (fry) assessments are
conducted, fry numbers in the 2005 brood year (11 million) were above the long-term cycle
average of 8 million (1980-2005).
Of the miscellaneous Early Summer Run stocks, Chilliwack Lake-Dolly Varden Creek (Upper
Chilliwack River) and the North Thompson River is reported separately in Table 1 due to their
high escapements in recent years. Chilliwack Lake-Dolly Varden Creek had a particularly high
escapement in 2004 (20,000 effective female spawners) and, subsequent returns in 2008 were
estimated at 83,000. The escapement for the 2005 brood year, however, was relatively low at
2,000 effective female spawners. Similarly, the North Thompson River has had high numbers of
effective female spawners in recent years with 47,000 sockeye escaping in the 2005 brood
year. Since there are no associated recruitment data for these two stocks, they are forecast
similarly to all other miscellaneous Early Summer Run stocks through the product of the brood
year effective female spawners and the average recruits per spawner for the eight stocks with
stock and recruitment data.
10
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Physical conditions (water levels and temperature) on the spawning grounds were conducive to
successful spawning. Nonetheless, the egg-to-fry survival rates (assuming average fecundity of
4,000 eggs/spawner) for the 2005 brood year of Nadina (38%) was below average (1973-2006:
47%).
Forecasts generated for each of these eight stocks produced similar forecast distributions for
the top three ranked models; the median forecast of one model was not more extreme than the
25% or 75% probability level forecast of another model. Although an alternative data set (fry
data) exists for Gates creek, no comparative fry models were run for this system given the
inconsistencies in assessment methods used to produce the fry data estimates. For Nadina,
the only other Early Summer Run stock with fry data, the top ranked fry model (ranked 4th)
produced similar results to the top three ranked models that rely on effective female spawner
data.
The 2009 median (50% probability level) return forecast for Early Summer Run sockeye is
739,000. Based on the 2009 forecast distribution there is a 25% probability the number of
returning sockeye will exceed 1,338,000 sockeye and a 75% probability that the return will
exceed 443,000 sockeye (Table 1). The 2009 forecast excluding the miscellaneous group (50%
probability forecast: 443,000) is above the long term average return for this cycle (1980-2005:
316,000).
Summer Run Sockeye
The Summer Run consists of four stocks: Chilko, Late Stuart, Quesnel and Stellako (Table 1).
The 2009 cycle is the dominant cycle for Late Stuart and Quesnel. Escapement in the 2005
brood year was 1.3 million effective female spawners for these four stocks which is below the
cycle average of 1.6 million (1980-2005). Quesnel comprised the largest percentage of this
total escapement (60%), followed by Chilko (20%), Late Stuart (10%), and Stellako (10%).
The number of effective female spawners in the 2005 brood year for Late Stuart and Quesnel
was respectively, 54% and 21% below their cycle average (1980-2005) (Table 3). For Chilko
and Stellako, the number of effective female spawners was respectively, 23% and 61% above
their cycle average (1980-2005) (Table 3).
Spawning success in the 2005 brood year was high; physical conditions on the spawning
grounds were conducive to successful spawning with water levels and temperature within an
acceptable range.
Chilko and Quesnel stocks have juvenile data (smolt and fry data respectively); no juvenile
assessments have been conducted for Stellako in recent years. Chilko juveniles are
enumerated as smolts as they migrate through an enumeration fence. Chilko smolt numbers
outmigrating in 2007 (2005 brood year) of 77 million were well above the cycle average of 23
million (1980-2005). Chilko smolt sizes in 2007 (88.4 mm) were also larger than average (82.9
mm " 5.69). Quesnel fry are enumerated using hydroaccoustics in Quesnel Lake in the fall prior
to their outmigration. Quesnel fry numbers in 2006 (2005 brood year) of 52 million were below
the cycle average of 58 million (1980-2005). Quesnel fry sizes in 2007 (2.7 g) were smaller
than the cycle average (3.6 g " 0.6) and similar in size to the 2001 brood year (2.6 g) when
survivals and consequently returns were poorer than average. These sizes, however, are not
as small as the 2002 brood year (1.9 g) when returns were well below average.
For Chilko and Late Stuart, forecasts are similar for the top three ranked models; the median
forecast of one model was not more extreme than the 25% or 75% probability level forecast of
11
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
another model. For Chilko, only smolt models were compared given the unprecedented large
number of smolts estimated for this stock relative to escapement. For Late Stuart, the top
ranked naïve RAC model (average return for the cycle) was ruled out given the brood year
escapement for this stock was half the cycle average and, therefore, the RAC forecast would
overestimate returns.
For Stellako, the model that performed best according to the
retrospective analysis (Ricker-cycle) was ruled out given the brood year escapement was above
all other escapements of the cycle line data set and when using the Ricker-cycle model
produced a lower forecast. The third ranked model for Stellako (R2C) was also ruled out given
the lower escapements in the past two cycle years relative to the 2005 brood year.
For Quesnel, top ranked models with environmental variables as covariates were ruled out.
These models produced forecasts that increased or decreased conversely to all other stock’s
biological models with the same environmental covariates. For example, including a sea
surface temperature covariate produced lower forecasts for Quesnel, despite the cooler than
average water temperatures in 2007 that should have increased marine survival, and therefore,
the forecast. The power (fry) and Larkin models ranked 6th after environmental models and
were used to generate the top forecasts. Given forecasts for these models diverged (the
median forecast of one model was more extreme than the 25% or 75% probability level forecast
of the other model), the two forecasts were averaged (weighted by retrospective performance).
The 2009 median (50% probability level) return forecast for Summer Run sockeye is 8,677,000.
Based on the 2009 forecast distribution there is a 25% probability the number of returning
sockeye will exceed 16,071,000 and a 75% probability that the return will exceed 4,914,000
sockeye (Table 1). The 2009 forecast (50% probability forecast: 8,677,000) is below the long
term average return for this cycle (1980-2005: 11,000,000).
Late Run Sockeye
The Late Run consists of six stocks: Cultus, Harrison, Late Shuswap, Portage, Weaver, and
Birkenhead (Table 3). The 2005 brood year is on an off cycle for the highly cyclic Late Shuswap
stock. Escapement in the 2005 brood year was 280,000 effective female spawners for these
stocks which is above the cycle average of 100,000 (1980-2005). In 2005, the total number of
effective female Late Run spawners, including the miscellaneous stocks was 290,000. The
spawning escapement for Harrison, in particular, has been well above documented
escapements in the 2005 (200,000 effective females) and 2006 (90,000 effective females)
brood years; Harrison is comprised of age-4 (41) fish and age-3 (31).
For all stocks except Cultus and Birkenhead, the brood year escapements were above their
1980-2005 cycle average (Table 3); Harrison (85%), Late Shuswap (72%), Portage (41%), and
Weaver (20%). Cultus had a particularly low brood year escapement (52 effective spawners:
112 adult spawners x 0.53 spawner success) at 1% of the cycle average (4,500 effective
spawners). However, due to hatchery supplementation of fry and smolts into the Cultus system,
the number of outmigrating smolts (98,000) was above the cycle average (1980-2005)(Table 3)
for this stock (50,000) and a smolt-jack model was used to forecast the number of returning
salmon for this stock. Almost all smolts outmigrating in 2007 (2009 returns) were hatchery
origin. Of the total outmigrating smolts, 90% were adipose-fin clipped indicating they were
hatchery origin and 10% were unclipped (adipose-fin present). The 10% unclipped fish likely
has a large component of hatchery-origin fish given unclipped hatchery fry were released into
the lake in 2006 (outmigrated as smolts in 2007); currently, it is not possible to determine the
proportion of outmigrating smolts that were wild versus hatchery origin.
12
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Weaver was the other stock where a juvenile salmon model was used to forecast returns. Fry
numbers (36 million) estimated for Weaver in 2006 (2005 brood year) were above the cycle
average of 26 million. Juvenile abundance (hydroacoustic estimates) for Late Shuswap
sockeye was not estimated in 2006.
Spawning success in the 2005 brood year was high; physical conditions on the spawning
grounds were conducive to successful spawning with water levels and temperature within an
acceptable range.
To forecast Harrison 2009 returns, an estimate of age-3 and age-4 returns for the 2005 brood
year was generated to produce a stock-recruit data point in the recent year’s high escapement
range; without this no data existed to inform the model at the recent year’s high escapements.
Age-3 returns in 2008 (2005 brood year) were estimated as a product of in-season return
estimated in 2008 (~46,000) and the proportion age-3’s in these returns (~9%). Age-4 returns
in 2009 (2005 brood year) were estimated based on the proportion of total recruits age-4’s
comprise using the age-3 returns in 2008 estimate. Using this extra data point in the Harrison
stock-recruit time series, forecasts were generated for the top ranked models for this stock.
The Harrison forecast has greater uncertainty compared to all other forecasts. In-season
Harrison run size estimate for 2008 used to estimate the 2005 brood year returns are
preliminary and the 2005 (age-4’s returning in 2009) and 2006 (age-3’s returning in 2009) brood
year escapements were estimated with lower accuracy/precision visual survey methods, despite
recent large escapements.
Forecasts generated for the top three ranked models for all six Late Run stocks produced
similar forecast distributions (see METHODS). For Cultus, naïve and adult models were
excluded from comparisons given the hatchery supplementation of this stock which resulted in
above average smolt outmigration despite low numbers of adult spawners in recent years. For
Harrison, given the well above average brood year escapements in 2005 and 2006, naïve
models were ruled out and only biological models were used. For Weaver, the top model using
fry data (ranked 4th) produced similar results compared to the top three ranked models based on
effective female spawner numbers. For Late Shuswap, the naïve model RAC (average return
across the cycles) was ruled out given the 2005 brood year escapement was approximately four
times the cycle average escapement making it likely for the RAC model to underestimate
returns.
The 2009 median (50% probability level) return forecast for Late Run sockeye is 907,000.
Based on the 2009 forecast distribution there is a 25% probability the number of returning
sockeye will exceed 1,616,000 sockeye and a 75% probability that the return will exceed
517,000 sockeye (Table 1). The 2009 forecast excluding the miscellaneous groups (50%
probability forecast: 843,000) is below the long term average return for this cycle (1980 to 2005:
1,036,000).
Pink Salmon
Fraser pink salmon are age-2 fish (21), spending their first winter in freshwater and their last
winter in the marine environment, prior to returning to the freshwater to spawn; Fraser pink
salmon immediately migrate to the ocean after their emergence from the spawning gravel. In
the Fraser River, pink salmon are an odd year run with negligible numbers migrating to the
system in even years. Therefore, pink salmon that return in 2009 are recruited from eggs
spawned by adults in 2007 (brood year). The total number of outmigrating fry in 2008 (2007
13
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
brood year) was 500 million, which is greater than the long term average of 400 million (19612007). Forecasts generated for the top ranked models produced similar forecast distributions;
the median forecast of one model was not more extreme than the 25% or 75% probability level
forecast of another model.
The 2009 median (50% probability level) return forecast for Fraser River pink salmon is 17.5
million. Based on the 2009 forecast distribution there is there is a 25% probability the number
of returning pink salmon will exceed 24.9 million and a 75% probability that the return will
exceed 12.5 million (Table 1). The 2009 forecast is above the long term average return for
Fraser River pink salmon (1961-2005: 12 million).
CONCLUSIONS
Forecasts are associated with relatively high uncertainty, consistent with previous Fraser
sockeye forecast PSARC reviews (Cass et al. 2006) and recent research on coast-wide salmon
stocks (Haeseker et al. 2007 & 2008). In attempts to improve forecast performance, individual
ocean indicators of salmon survival (e.g. sea-surface temperature, Fraser discharge, etc.) have
been used as covariates in forecast models and compared retrospectively. However,
environmental variables explored to date have not explained a significant component of salmon
survival or recruitment variability (Myers 1998; Mueter et al. 2005; Cass et. al 2006) and
individual indicators alone have also not been able to explain the extremely poor ocean survival
conditions encountered by sockeye salmon that migrated to the ocean in 2005 and returned in
poor numbers (well below forecast) in 2007. In a recent study that explored systemic temporal
shifts in productivity (annual Ricker-a parameter estimates) between salmon species and
stocks, no consistent productivity trends were identified specifically for Fraser sockeye stocks
(Dorner et al. 2008). These results suggest that the survival responses of Fraser sockeye to
environmental and other factors are complex and, at present, not well understood (Dorner et al.
2008).
Wild Salmon Policy Conservation Units
To protect biological diversity of wild salmon stocks, the first strategy described by the Wild
Salmon Policy (DFO 2005) includes the identification and inventory of the units of diversity to
conserve. The methodology for the identification of Pacific wild salmon conservation units
(CU’s) and the CU’s are reported in Holtby & Ciruna (2007). For future conservation and
management of salmon resources, the CU’s that each forecasted stock comprises is
numerically referenced from 1 to 32 in Table 1, with the corresponding CU name and index
listed in Table 2.
Ocean Conditions
To provide some indication of survival conditions for Fraser River sockeye salmon, a
compilation of ocean survival indicators for salmon are presented (Table 5) to qualitatively
compare relative ocean survival conditions from 1998 to 2007. Annual ocean indicators are
linked to the period sockeye salmon first migrate to the ocean (e.g. 2007 for most sockeye
returning in 2009), when salmon are at their smallest size in their ocean phase and, therefore,
most vulnerable to size-dependent mortality.
14
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
The methodology for ranking individual indicators is based on W.T. Peterson’s (U.S. Northwest
Fisheries Sciences Centre, National Ocean and Atmospheric Agency) approach:
http://www.nwfsc.noaa.gov/research/divisions/fed/oeip/g-forecast.cfm. Currently, the suite of
ocean indicators explored only partially track Chilko marine survival (marine survival indicator
system for Fraser sockeye stocks) (Table 5). This suggests that more ocean indicators need to
be explored and/or developed to improve forecasting methodology. The suite of indicators
currently used (Table 5) does provide a strong indication of the very poor ocean survival
conditions experienced by Fraser sockeye in 2005 (poor returns in 2007). Therefore, it may
provide an indication of below average returns. For sockeye returning in 2009, ocean
conditions in their juvenile outmigration period (spring/summer 2007) were good for
most physical and biological indicators (Table 5). Although emphasis on median
probability levels (50%) is recommended for 2009 sockeye return forecasts, further work
is required for ocean indicators of sockeye survival. The indicators of ocean condition
are as follows:
1. Pacific Decadal Oscillation (PDO): a broad index of sea surface temperature in the North
Pacific (Mantua et al. 1997); http://jisao.washington.edu/pdo/PDO.latest
2. Aleutian Low Pressure Index (ALPI): an index of the relative intensity of the Aleutian Low
pressure system of the North Pacific in the winter period (Beamish et al. 1997);
http://www.pac.dfo-mpo.gc.ca/sci/sa-mfpd/downloads/indices/alpi.txt
3. Sea-Surface-Temperature (SST) Entrance Island: average SST data (April to June) in the
SOG where juvenile Fraser sockeye first enter the marine environment. http://www.pac.dfompo.gc.ca/sci/OSAP/data/SearchTools/Searchlighthouse_e.htm
4. Sea-Surface-Temperature (SST) Pine Island: average SST data (April to July) on the
northern tip of Vancouver Island (see previous weblink).
5. Coastal Upwelling Index (CUI): an index of the coastal water volume that upwells (Bakun
1973);
can
provide
an
indication
of
coastal
ocean
productivity.
http://www.pfeg.noaa.gov/products/PFEL/modeled/indices/upwelling/NA/upwell_menu_NA.html
6. Physical Spring Transition: the period in the coastal ocean when the winter downwelling
period transitions to a summer upwelling period. Generally the earlier this transition the more
productive the coastal ocean (See previous (CUI) weblink).
7. Southern Copepods (SVI: Southern Vancouver Island & NVI: Northern Vancouver
Island): typically occur in more southern latitudes (e.g. coastal California); their increased
abundance in coastal BC waters provides an indication of warmer ocean conditions; southern
copepods are smaller and have a lower energetic content compared to boreal shelf copepods
(data obtained from D.L. Mackas, Institute of Ocean Sciences, Canadian Department of
Fisheries and Oceans; see Irvine & Crawford 2008).
8. Boreal Shelf Copepods (SVI: Southern Vancouver Island & NVI: Northern Vancouver
Island): typically occur in BC coastal waters; their abundance in coastal BC waters provides an
indication of average to cooler ocean conditions; boreal shelf copepods are larger and have a
higher energetic content (see previous reference).
15
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Table 5. Indicators of ocean conditions (1998-2007) (methodology & approach from W.T. Peterson, U.S.
NFSC, NOAA). For 2009 returns: most sockeye (age-4’s) spawned in 2005 and migrated to the ocean in
2007. All data used is referenced above. For each indicator, annual estimates were ranked across all
years from 1 to 10 from best to worst salmon ocean survival conditions. Green (G): ranks 1 to 4; yellow
(Y): ranks 5 to 7; red (R): ranks 7 to 10.
(BROOD YEAR)
OCEAN ENTRY YEAR
(1996)
(1997)
(1998)
(1999)
(2000)
(2001)
(2002)
(2003)
(2004)
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
(RETURN YEAR)
(2000)
(2001)
(2002)
(2003)
(2004)
(2005)
(2006)
(2007)
(2008)
(2009)
G
Y
G
G
R
Y
G
R
1 PDO (Jan-March average)
R
G
G
R
G
R
R
R
Y
Y
2 ALPI
R
G
Y
R
R
R
R
Y
G
G
3 SST (Entrance Island)
R
G
G
G
G
R
R
R
Y
Y
4 SST (Pine Island)
R
G
G
G
Y
R
R
R
Y
G
Chilko Marine Survival
(2005)
NA
NA
Ocean Indices
Physical Conditions
5 Upwelling index (48ºN)
G
G
R
Y
G
R
Y
R
Y
G
6 Spring transition timing (48ºN)
G
G
Y
Y
G
Y
Y
R
Y
Y
Biological Conditions
7 Southern Copepods (SVI)
R
G
Y
G
G
R
Y
R
R
G
8 Boreal Shelf Copepods (SVI)
R
G
G
Y
G
Y
R
R
R
G
9 Southern Copepods (NVI)
R
G
G
G
Y
R
Y
R
R
G
Y
G
G
R
G
R
R
R
Y
G
10 Boreal Shelf Copepods (NVI)
SOURCES OF INFORMATION
Bakun, A. 1973. Coastal upwelling indices, west coast of North America, 1946–71.
Department of Commerce, NOAA Technical Report NMFS–SSRF–671.
U.S.
Beamish, R.J., Neville, C.E., & Cass, A.J. 1997. Production of Fraser River sockeye salmon
(Oncorhynchus nerka) in relation to decadal-scale changes in the climate and the ocean.
Can. J. Fish. Aquat. Sci. 54: 543-554.
Cass, A.J., Folkes, M., Parken, C.K., & Wood, C.C. 2006. Pre-season run size f orecasts
Fraser River sockeye for 2006. PSARC Working Paper S2006/060.
for
DFO. 2005. Canada’s policy for conservation of wild Pacific salmon. Fisheries & Oceans
Canada, 401 Burrard St., Vancouver, BC, V6C 3S4. p. 49#v.
DFO. 2007. Pre-season run size forecasts for Fraser River sockeye and pink salmon in 2008.
DFO Can. Sci. Advis. Rep. 2007/049.
DFO. 2008. State of the Pacific Ocean 2007. DFO Can. Sci. Advis. Rep. 2008/028
Dorner, B., Peterman, R.M., & Haeseker, S.L. 2008. Historical trends in productivity of 120
Pacific pink, chum and sockeye salmon stocks reconstructed by using a Kalman filter.
Can. J. Fish. Aquat. Sci. 65: 1842-1866.
16
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
Haeseker, S.L., Dorner, B., Peterman, R.M., & Zhenming, S. 2007. An improved sibling model
for forecasting chum salmon and sockeye salmon abundance. N. Am. J. Fish. Manag.
27: 634-642.
Haeseker, S.L., Peterman, R.M., & Zhenming, S. 2008. Retrospective evaluation of preseason
forecasting models for sockeye and chum salmon. N. Am. J. Fish. Manag. 28: 12-29.
Irvine, J. & Crawford, W. (editors). 2008. State of physical, biological, and selected fishery
resources of Pacific Canadian marine ecosystems. CSAS Res. Doc. 2008/013. 109 pp.
Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., & and Francis, R.C. 1997. A Pacific
decadal climate oscillation with impacts on salmon. Bull. Am. Meterol. Soc. 78: 10691079.
Mueter, F.J., Pyper, B.J., & Peterman, R.M. 2005. Relationship between coastal ocean
conditions and survival rates of Northeast Pacific Salmon at multiple lags. Trans. Am.
Fish. Soc. 134: 105-119.
Myers, R.A. 1998. When do environment-recruitment correlations work? Rev. Fish. Bio. Fish.
8: 285-305.
17
Pacific Region
Pre-season run size forecasts for
Fraser River sockeye & pink salmon in 2009
FOR MORE INFORMATION
Contacts:
Tel:
Fax:
E-Mail:
Sue Grant
Fraser River Stock Assessment
Fisheries and Oceans Canada
100 Annacis Parkway, Unit 3
Delta, BC V3M 6A2
604-666-7270
604-666-7112
Sue.Grant@dfo-mpo.gc.ca
Alan Cass
Pacific Biological Station
Fisheries and Oceans Canada
3190 Hammond Bay Road
Nanaimo, BC V9T 6N7
250-756-7142
250-756-7209
Alan.Cass@dfo-mpo.gc.ca
This report is available from the:
Centre for Science Advice (CSA)
Pacific Region
Fisheries and Oceans Canada
3190 Hammond Bay Road
Nanaimo, BC V9T 6N7
Telephone:250-756-7208
Fax:250-756-7209
E-Mail: psarc@dfo-mpo.gc.ca
Internet address: www.dfo-mpo.gc.ca/csas
ISSN 1919-5079 (Print)
ISSN 1919-5087 (Online)
© Her Majesty the Queen in Right of Canada, 2009
La version française est disponible à l’adresse ci-dessus.
CORRECT CITATION FOR THIS PUBLICATION
DFO. 2009. Pre-season run size forecasts for Fraser River sockeye and pink salmon in 2009.
DFO Can. Sci. Advis. Sec. Sci. Advis. Rep. 2009/022
18
Download